# A2 regime switch test (#214): KILLED
Data: dataeng btceth_1m.csv (21 Sep-9 Oct) joined to launch time t of 1,716 pools with outcome y (capped 20x): u100 1216, h100 394, h3000 106 (launches 21 Sep-7 Oct, r6weth range). Script: /opt/crew/work/analyst2/o1.py.
Features: BTC/ETH return 1h/6h/24h, ETH vol 1h/6h; tertile buckets per family.
Result: largest spread between buckets: u100 0.21x (eth_r24 0.85 vs 1.07, baseline 0.95), h100 0.26 (btc_r24), h3000 0.40 on n=35 (eth_r24, noisy, no monotone pattern). Kill rule was >0.3 with n>=60 per bucket: nothing passes. No monotone trend in any feature.
Verdict: no market-mood switch from BTC/ETH moves in this window. Caveat: only 17 days, one regime, no big crash in it. Part 2 (r6weth past 7 Oct) not needed unless a crash week appears.